Linking and reanalysis of educational experiments with administrative data: Findings and impact

ObjectivesThis project combined administrative data with completed randomised controlled trials to understand whether educational interventions led to additional benefits or costs beyond those initially evaluated. The purpose of the project was to demonstrate that this can be done, the value of such work, and what needs to happen next to embed such approaches.
 MethodsWe reviewed the Education Endowment Foundation trials database for studies that could plausibly link to other outcomes: exclusion / suspension from school, criminal conviction, or being Not in Education, Employment or Training (NEET). After shortlisting we intended to link at least one trial to different administrative datasets containing these outcomes.
 We undertook three other tasks:
 
 Understanding barriers to data sharing within government from a behavioural perspective and proposed solutions to those challenges.
 Reviewing theories of change for some interventions to assess whether/how there could be pathways to the outcomes we were looking at.
 Asking intervention developers and evaluators for their predictions and rationales for possible impacts on other outcomes, without them kn